Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:129966
Precedence: list
MIME-Version: 1.0
References: <CAPyj-LC0fn2NvQEpebo6NbL5R8svZ8CmtHUhb5BkUE4W49eXzw@mail.gmail.com>
 <AM8P250MB0170DD4D1FB2EC74D6D1901EE29CA@AM8P250MB0170.EURP250.PROD.OUTLOOK.COM>
In-Reply-To: <AM8P250MB0170DD4D1FB2EC74D6D1901EE29CA@AM8P250MB0170.EURP250.PROD.OUTLOOK.COM>
Date: Sat, 31 Jan 2026 17:32:48 +0100
Message-ID: <CAPyj-LCED3O-GdnrWBeZY+AR+mSzbE+WaM53oj99YHzkpwAX0w@mail.gmail.com>
Subject: Re: [PHP-DEV] [RFC][Discussion] Closure optimizations
To: PHP internals <internals@lists.php.net>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
From: tovilo.ilija@gmail.com (Ilija Tovilo)

Hi Bob

On Sat, Jan 31, 2026 at 4:59=E2=80=AFAM Bob Weinand <bobwei9@hotmail.com> w=
rote:
>
> On 31.1.2026 00:40:05, Ilija Tovilo wrote:
>
> https://wiki.php.net/rfc/closure-optimizations
>
> I'm a bit disappointed that you felt necessary to introduce bureaucracy f=
or a straightforward very minor BC break. I don't like this something to be=
come necessary or expected in future. Sending that email was completely suf=
ficient.
>
> I would like to encourage you to recognize that and simply merge your pul=
l request.

I agree that the BC break is fairly minimal. What made me decide to
create an RFC after all is highlighted here:

> ReflectionFunction::getClosureThis() will now return NULL for closures th=
at are inferred as static. This might be slightly surprising, given static =
inference does not work 100% reliably, limited by the rules previously desc=
ribed.

ReflectionFunction::getClosureThis() will work as long as the closure
uses $this, but it will also work if it calls static methods. Once you
remove the static call, it will break. Now, I don't think
ReflectionFunction::getClosureThis() is something many people rely on,
nor should they, but it is nonetheless surprising. I'd like to avoid
the situation where this is categorized as a bug that is effectively
impossible to solve without removing the optimization.

Ilija